| Server IP : 68.65.120.201 / Your IP : 216.73.216.184 Web Server : LiteSpeed System : Linux server179.web-hosting.com 4.18.0-513.18.1.lve.el8.x86_64 #1 SMP Thu Feb 22 12:55:50 UTC 2024 x86_64 User : taxhyuvu ( 2294) PHP Version : 8.1.33 Disable Function : NONE MySQL : OFF | cURL : ON | WGET : ON | Perl : ON | Python : ON | Sudo : OFF | Pkexec : OFF Directory : /opt/cloudlinux/venv/lib64/python3.11/site-packages/cl_website_collector/ |
Upload File : |
# -*- coding: utf-8 -*-
#
# Copyright © Cloud Linux GmbH & Cloud Linux Software, Inc 2010-2024 All Rights Reserved
#
# Licensed under CLOUD LINUX LICENSE AGREEMENT
# http://cloudlinux.com/docs/LICENSE.TXT
import json
import logging
import os
import time
from datetime import datetime, timezone
from pathlib import Path
from typing import Dict, Any, Optional
import requests
from clcommon.cpapi import cpusers, userdomains
from cl_website_collector.apache_processor import ApacheProcessor
from cl_website_collector.docroot_processor import DocrootProcessor
from cl_website_collector.constants import (
VERSION,
STORAGE_DIR,
OPT_OUT_FILE,
DOCROOT_SCAN_TIMEOUT,
DOCROOT_LIMIT_PER_RUN,
SENDING_TIMEOUT,
MAX_RECORD_SIZE,
FILE_SIZE_LIMIT,
)
class WebsiteCollector:
"""
Website data collection.
Collects and exports various Apache-related data:
- Apache configuration files (.conf) from standard directories
- Apache system information (version, loaded modules)
- .htaccess files from website document roots
- Maintains processing state to avoid duplicate scans
"""
def __init__(self, logger: logging.Logger, storage_dir: Optional[Path] = None):
"""
Initialize website collector for on-demand data collection and sending.
Args:
storage_dir: Directory for storing website collector state
"""
# Use default directories if not provided
if storage_dir is None:
storage_dir = STORAGE_DIR
self.logger = logger
# Ensure directories exist
self._ensure_directories(storage_dir=storage_dir)
# Initialize components for on-demand data collection
self.docroot_processor = DocrootProcessor(logger=self.logger)
self.apache_processor = ApacheProcessor(logger=self.logger)
# Initialize sending stats tracking
self.stats_file = storage_dir / "website_collector_stats.json"
self.dry_run_file = storage_dir / "dry-run.jsonl"
self.sending_stats = {'global_apache_configs_sent': False, 'docroots': {}} # default values
# Load sending stats
self._load_sending_stats()
def _get_docroots_for_sending(self) -> Dict:
"""
Get docroots for sending, with optional filtering and limit.
Returns:
Dict of docroots ready for processing (limited and filtered by sent status)
"""
selected_docroots = {}
processed_count = 0
try:
# Use Panel API, but optimize by checking processed status early
panel_users = cpusers()
for username in panel_users:
try:
user_domains = userdomains(username)
user_docroots_found = {}
# Collect ALL domains for ALL docroots for this user first
for domain, docroot in user_domains:
docroot_path = Path(docroot)
if docroot_path.exists() and os.access(str(docroot_path), os.R_OK):
# Check if already fully sent - skip processing
if self.is_docroot_fully_sent(docroot):
processed_count += 1
continue
# Add to user's docroots (collect ALL domains per docroot)
if docroot not in user_docroots_found:
user_docroots_found[docroot] = {
"docroot": docroot,
"username": username,
"domains": []
}
user_docroots_found[docroot]["domains"].append(domain)
# Add user's docroots to global selection (check limit at docroot level)
limit_reached = False
for docroot, docroot_data in user_docroots_found.items():
if len(selected_docroots) >= DOCROOT_LIMIT_PER_RUN:
limit_reached = True
break
selected_docroots[docroot] = docroot_data
# If we reached the limit, stop processing more users
if limit_reached:
break
except Exception as e:
self.logger.debug("Failed to get domains for user %s: %s", username, e)
except Exception as e:
self.logger.error("[WEBSITE-COLLECTOR] Panel API failed: %s", e)
self.logger.debug("Found %d unprocessed docroots (skipped %d already processed)",
len(selected_docroots), processed_count)
return selected_docroots
def _load_sending_stats(self) -> None:
"""Load sending statistics from tracking file."""
try:
if self.stats_file.exists():
with open(self.stats_file, 'r', encoding='utf-8') as f:
self.sending_stats = json.load(f)
self.logger.debug("Loaded sending stats")
except Exception as e:
self.logger.debug("Failed to load sending stats: %s", e)
def read_file_content(self, file_path: str) -> Optional[str]:
"""
Read any file content on-demand for sending.
Supports both Apache configs and .htaccess files.
"""
try:
p = Path(file_path)
is_symlink = p.is_symlink()
real_p = p.resolve(strict=False) if is_symlink else p
if not (real_p.exists() and real_p.is_file() and os.access(str(real_p), os.R_OK)):
self.logger.error("[WEBSITE-COLLECTOR] Cannot read file: %s", file_path)
return None
if real_p.stat().st_size > FILE_SIZE_LIMIT:
self.logger.error("[WEBSITE-COLLECTOR] File too big: %s (size: %d bytes)", file_path,
real_p.stat().st_size)
return None
return real_p.read_text(encoding='utf-8', errors='replace')
except Exception as e:
self.logger.error("[WEBSITE-COLLECTOR] Error reading file %s: %s", file_path, e)
return None
def _save_sending_stats(self) -> None:
"""Save sending statistics to tracking file."""
try:
self.sending_stats['last_updated'] = datetime.now(timezone.utc).isoformat()
with open(self.stats_file, 'w', encoding='utf-8') as f:
json.dump(self.sending_stats, f, indent=2, ensure_ascii=False)
self.logger.debug("Saved sending stats")
except Exception as e:
self.logger.error("[WEBSITE-COLLECTOR] Failed to save sending stats: %s", e)
def is_docroot_fully_sent(self, docroot: str) -> bool:
"""Check if docroot data was fully sent (both htaccess and user apache configs)."""
docroot_stats = self.sending_stats.get('docroots', {}).get(docroot, {})
return (docroot_stats.get('htaccess_sent', False) and
docroot_stats.get('user_apache_configs_sent', False))
def mark_docroot_htaccess_sent(self, docroot: str, username: str) -> None:
"""Mark htaccess data as sent for docroot."""
if docroot not in self.sending_stats['docroots']:
self.sending_stats['docroots'][docroot] = {
'htaccess_sent': False,
'user_apache_configs_sent': False,
'username': username
}
self.sending_stats['docroots'][docroot]['htaccess_sent'] = True
self.sending_stats['docroots'][docroot]['last_processed'] = datetime.now(timezone.utc).isoformat()
def mark_docroot_apache_sent(self, docroot: str, username: str) -> None:
"""Mark user apache configs as sent for docroot."""
if docroot not in self.sending_stats['docroots']:
self.sending_stats['docroots'][docroot] = {
'htaccess_sent': False,
'user_apache_configs_sent': False,
'username': username
}
self.sending_stats['docroots'][docroot]['user_apache_configs_sent'] = True
self.sending_stats['docroots'][docroot]['last_processed'] = datetime.now(timezone.utc).isoformat()
def mark_global_apache_sent(self) -> None:
"""Mark global apache configs as sent."""
self.sending_stats['global_apache_configs_sent'] = True
@staticmethod
def is_collection_disabled() -> bool:
"""
Check if website collection is disabled via opt-out file.
Returns:
True if collection is disabled, False otherwise
"""
return OPT_OUT_FILE.exists()
def _ensure_directories(self, storage_dir: Path) -> None:
"""
Create necessary directory structure.
Args:
storage_dir: Directory for storing website collector state
"""
storage_dir.mkdir(mode=0o750, parents=True, exist_ok=True)
def _create_base_record(self, client_id: str, system_id: str, platform: str, panel: str) -> Dict[str, Any]:
"""Create base record with common fields."""
return {
'collection_timestamp': int(time.time()),
'collection_version': VERSION,
'client_id': client_id,
'system_id': system_id,
'platform': platform,
'panel': panel,
'data': {},
'meta_data': {}
}
def _estimate_record_size(self, record: Dict[str, Any]) -> int:
"""
Estimate record size in JSON bytes.
Returns:
Estimated size in bytes
"""
return len(json.dumps(record, ensure_ascii=False).encode('utf-8'))
def _create_new_record_part(self, base_record: Dict[str, Any]) -> Dict[str, Any]:
"""
Create new record part with base data but empty data/meta_data sections.
Args:
base_record: Base record to copy from
Returns:
New record with empty data sections
"""
new_record = base_record.copy()
new_record['data'] = {}
new_record['meta_data'] = {}
return new_record
def _add_file_to_records(self, file_path: str, file_content: str, is_symlink: bool,
current_record: Dict[str, Any], base_record: Dict[str, Any],
records_to_send: list) -> Dict[str, Any]:
"""
Add file to current record, splitting if size limit exceeded.
Args:
file_path: Path of the file to add
file_content: Content of the file
is_symlink: Whether the file is a symlink
current_record: Current record being built
base_record: Base record template for new parts
records_to_send: List to append completed records to
Returns:
Updated current record (may be new if split occurred)
"""
# Create temporary copy to check size
temp_record = current_record.copy()
temp_record['data'] = current_record['data'].copy()
temp_record['meta_data'] = current_record['meta_data'].copy()
# Add new file to temporary copy
temp_record['data'][file_path] = file_content
if is_symlink:
temp_record['meta_data'][file_path] = {'is_symlink': True}
# Check size
estimated_size = self._estimate_record_size(temp_record)
if estimated_size > MAX_RECORD_SIZE and current_record['data']:
# Current record is full, save it and create new one
records_to_send.append(current_record)
current_record = self._create_new_record_part(base_record)
# Add file to new record
current_record['data'][file_path] = file_content
if is_symlink:
current_record['meta_data'][file_path] = {'is_symlink': True}
else:
# File fits, use temporary copy
current_record = temp_record
return current_record
def _send_or_collect_record(self, record: Dict[str, Any], session, api_url: str, dry_run: bool) -> bool:
"""Send record or save to JSONL file for dry run. Returns True if successful."""
if dry_run:
try:
# Save to dry-run.jsonl file (append mode)
with open(self.dry_run_file, 'a', encoding='utf-8') as f:
f.write(json.dumps(record, ensure_ascii=False, separators=(',', ':')) + '\n')
return True
except Exception as e:
self.logger.error("[WEBSITE-COLLECTOR] Error saving dry-run record: %s", e)
return False
else:
# Try sending with retries
max_retries = 3
last_error = None
last_status_code = None
for attempt in range(max_retries):
try:
response = session.post(api_url, json=record, timeout=30)
if response.status_code == 200:
return True
last_status_code = response.status_code
if attempt < max_retries - 1:
self.logger.debug("Non-200 response %s for docroot %s, attempt %d/%d",
response.status_code, record.get('document_root', 'global'),
attempt + 1, max_retries)
# Exponential backoff: 1s, 2s, 4s
time.sleep(2 ** attempt)
else:
self.logger.debug("Final attempt failed: Non-200 response %s for docroot %s",
response.status_code, record.get('document_root', 'global'))
except Exception as e:
last_error = str(e)
if attempt < max_retries - 1:
self.logger.debug("HTTP error on attempt %d/%d for docroot %s: %s",
attempt + 1, max_retries, record.get('document_root', 'global'), e)
# Exponential backoff: 1s, 2s, 4s
time.sleep(2 ** attempt)
else:
self.logger.error("[WEBSITE-COLLECTOR] Final HTTP error while sending record: %s", e)
# All retries failed - determine failure reason
if last_status_code:
reason = f"HTTP {last_status_code}"
elif last_error:
reason = f"Exception: {last_error}"
else:
reason = "Unknown error"
self.logger.error("[WEBSITE-COLLECTOR] Failed to send record for docroot %s after %d retries. Reason: %s",
record.get('document_root', 'global'), max_retries, reason)
return False
def _send_global_configs(self, session, api_url: str, dry_run: bool,
client_id: str, system_id: str, platform: str, panel: str) -> bool:
"""
Send global Apache configs and system info, splitting into multiple records if needed.
Returns:
True if all records sent successfully, False otherwise
"""
base_record = self._create_base_record(client_id, system_id, platform, panel)
base_record.update({
'username': '',
'domains': [],
'document_root': '',
})
current_record = self._create_new_record_part(base_record)
records_to_send = []
# Collect global Apache config files
global_configs = self.apache_processor.collect_global_config_paths()
for path_info in global_configs.get('config_paths', []):
file_content = self.read_file_content(path_info['file_path'])
if file_content is None:
file_content = '# ERROR: File unreadable'
current_record = self._add_file_to_records(
path_info['file_path'], file_content, path_info.get('is_symlink', False),
current_record, base_record, records_to_send
)
# Add system info to the last record
system_info = self.apache_processor.get_apache_system_info()
current_record['data']['modules'] = system_info.get('loaded_modules', [])
current_record['data']['version'] = system_info.get('apache_version', '')
# Add last record
records_to_send.append(current_record)
# Send all records
for i, record in enumerate(records_to_send):
self.logger.debug("Sending global config record %d/%d", i + 1, len(records_to_send))
if not self._send_or_collect_record(record, session, api_url, dry_run):
return False
return True
def _send_docroot_data(self, docroot: str, domains: list, username: str, session, api_url: str,
dry_run: bool, client_id: str, system_id: str,
platform: str, panel: str) -> bool:
"""
Send all docroot data, splitting into multiple records if needed.
Returns:
True if all records sent successfully, False otherwise
"""
docroot_stats = self.sending_stats.get('docroots', {}).get(docroot, {})
base_record = self._create_base_record(client_id, system_id, platform, panel)
base_record.update({
'username': username,
'domains': domains,
'document_root': docroot,
})
current_record = self._create_new_record_part(base_record)
records_to_send = []
# Collect htaccess files
if not docroot_stats.get('htaccess_sent', False):
htaccess_data = self.docroot_processor.collect_htaccess_paths(docroot, domains, username,
timeout=DOCROOT_SCAN_TIMEOUT)
if htaccess_data and htaccess_data.get('htaccess_file_paths'):
for htaccess_path_info in htaccess_data['htaccess_file_paths']:
content = self.read_file_content(htaccess_path_info['real_path'])
if content is None:
content = '# ERROR: File unreadable'
current_record = self._add_file_to_records(
htaccess_path_info['file_path'], content, htaccess_path_info.get('is_symlink', False),
current_record, base_record, records_to_send
)
# Collect user Apache configs
if not docroot_stats.get('user_apache_configs_sent', False):
user_configs = self.apache_processor.collect_user_specific_config_paths([username])
if user_configs and user_configs.get('config_paths'):
for path_info in user_configs['config_paths']:
file_content = self.read_file_content(path_info['file_path'])
if file_content is None:
file_content = '# ERROR: File unreadable'
current_record = self._add_file_to_records(
path_info['file_path'], file_content, path_info.get('is_symlink', False),
current_record, base_record, records_to_send
)
# Add last record if it has data or if no records were created
if current_record['data'] or not records_to_send:
records_to_send.append(current_record)
# Send all records
for i, record in enumerate(records_to_send):
self.logger.debug("Sending docroot %s record %d/%d", docroot, i + 1, len(records_to_send))
if not self._send_or_collect_record(record, session, api_url, dry_run):
return False
return True
def send_data(self, system_id: str, client_id: str, platform: str, panel: str, api_url: str,
dry_run: bool = False) -> None:
"""
Send data directly or save to JSONL file in dry-run mode.
Read htaccess and apache configs on-demand and track sending status.
Args:
dry_run: If True, saves data to dry-run.jsonl file instead of sending to API
"""
start_time = time.time()
session = requests.Session() if not dry_run else None
try:
# Step 1: Send global Apache configs if not sent yet
if not self.sending_stats.get('global_apache_configs_sent', False):
self.logger.debug("%s global Apache configs", "Collecting" if dry_run else "Sending")
if self._send_global_configs(session, api_url, dry_run, client_id, system_id, platform, panel):
if not dry_run:
self.mark_global_apache_sent()
self._save_sending_stats()
self.logger.debug("Global Apache configs sent successfully")
# Step 2: Process docroots
unprocessed_docroots = self._get_docroots_for_sending()
for docroot, domain_data in unprocessed_docroots.items():
if time.time() - start_time > SENDING_TIMEOUT:
self.logger.error("[WEBSITE-COLLECTOR] Timeout reached during docroot processing")
break
self.logger.debug("Processing docroot %s for user %s", docroot, domain_data['username'])
if self._send_docroot_data(docroot, domain_data['domains'], domain_data['username'], session, api_url,
dry_run, client_id, system_id, platform, panel):
if not dry_run:
self.mark_docroot_htaccess_sent(docroot, domain_data['username'])
self.mark_docroot_apache_sent(docroot, domain_data['username'])
self._save_sending_stats()
self.logger.debug("Docroot %s sent successfully", docroot)
execution_time = time.time() - start_time
if dry_run:
self.logger.info("DRY RUN completed: saved to %s in %.1f seconds", self.dry_run_file, execution_time)
else:
self.logger.debug("Sequential sending completed: %.1f seconds", execution_time)
except Exception as e:
self.logger.error("[WEBSITE-COLLECTOR] Failed during %s: %s",
"dry run" if dry_run else "sequential sending",
e)
finally:
if session:
session.close()